curl_sac.py

import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
import copy
from matplotlib import pyplot as plt
import time
import math

from curl import utils
from curl.encoder import make_encoder

LOG_FREQ = 10000

def get_optimizer_stats(optim):
    state_dict = optim.state_dict()['state']
    if len(state_dict) ==0:
        flattened_exp_avg = flattened_exp_avg_sq = 0.
        print('Warning: optimizer dict is empty!')
    else:
        # flattened_exp_avg = flattened_exp_avg_sq = 0.
        flattened_exp_avg = torch.cat([x['exp_avg'].flatten() for x in state_dict.values()]).cpu().numpy()
        flattened_exp_avg_sq = torch.cat([x['exp_avg_sq'].flatten() for x in state_dict.values()]).cpu().numpy()
    return {
        'exp_avg_mean': np.mean(flattened_exp_avg),
        'exp_avg_std': np.std(flattened_exp_avg),
        'exp_avg_min': np.min(flattened_exp_avg),
        'exp_avg_max': np.max(flattened_exp_avg),
        'exp_avg_sq_mean': np.mean(flattened_exp_avg_sq),
        'exp_avg_sq_std': np.std(flattened_exp_avg_sq),
        'exp_avg_sq_min': np.min(flattened_exp_avg_sq),
        'exp_avg_sq_max': np.max(flattened_exp_avg_sq),
    }


def gaussian_logprob(noise, log_std):
    """Compute Gaussian log probability."""
    residual = (-0.5 * noise.pow(2) - log_std).sum(-1, keepdim=True)
    return residual - 0.5 * np.log(2 * np.pi) * noise.size(-1)


def squash(mu, pi, log_pi):
    """Apply squashing function.
    See appendix C from https://arxiv.org/pdf/1812.05905.pdf.
    """
    mu = torch.tanh(mu)
    if pi is not None:
        pi = torch.tanh(pi)
    if log_pi is not None:
        log_pi -= torch.log(F.relu(1 - pi.pow(2)) + 1e-6).sum(-1, keepdim=True)
    return mu, pi, log_pi


def weight_init(m):
    """Custom weight init for Conv2D and Linear layers."""
    if isinstance(m, nn.Linear):
        nn.init.orthogonal_(m.weight.data)
        m.bias.data.fill_(0.0)
    elif isinstance(m, nn.Conv2d) or isinstance(m, nn.ConvTranspose2d):
        # delta-orthogonal init from https://arxiv.org/pdf/1806.05393.pdf
        assert m.weight.size(2) == m.weight.size(3)
        m.weight.data.fill_(0.0)
        m.bias.data.fill_(0.0)
        mid = m.weight.size(2) // 2
        gain = nn.init.calculate_gain('relu')
        nn.init.orthogonal_(m.weight.data[:, :, mid, mid], gain)


class Actor(nn.Module):
    """MLP actor network."""

    def __init__(
      self, obs_shape, action_shape, hidden_dim, encoder_type,
      encoder_feature_dim, log_std_min, log_std_max, num_layers, num_filters
    ):
        super().__init__()

        self.encoder = make_encoder(
            encoder_type, obs_shape, encoder_feature_dim, num_layers,
            num_filters, output_logits=True
        )

        self.log_std_min = log_std_min
        self.log_std_max = log_std_max

        self.trunk = nn.Sequential(
            nn.Linear(self.encoder.feature_dim, hidden_dim), nn.ReLU(),
            nn.Linear(hidden_dim, hidden_dim), nn.ReLU(),
            nn.Linear(hidden_dim, 2 * action_shape[0])
        )

        self.outputs = dict()
        self.apply(weight_init)

    def forward(
      self, obs, compute_pi=True, compute_log_pi=True, detach_encoder=False
    ):
        obs = self.encoder(obs, detach=detach_encoder)

        mu, log_std = self.trunk(obs).chunk(2, dim=-1)

        # constrain log_std inside [log_std_min, log_std_max]
        log_std = torch.tanh(log_std)
        log_std = self.log_std_min + 0.5 * (
          self.log_std_max - self.log_std_min
        ) * (log_std + 1)

        self.outputs['mu'] = mu
        self.outputs['std'] = log_std.exp()

        if compute_pi:
            std = log_std.exp()
            noise = torch.randn_like(mu)
            pi = mu + noise * std
        else:
            pi = None
            entropy = None

        if compute_log_pi:
            log_pi = gaussian_logprob(noise, log_std)
        else:
            log_pi = None

        mu, pi, log_pi = squash(mu, pi, log_pi)

        return mu, pi, log_pi, log_std

    def forward_from_feature(self, obs, compute_pi=False, compute_log_pi=False):
        mu, log_std = self.trunk(obs).chunk(2, dim=-1)

        # constrain log_std inside [log_std_min, log_std_max]
        log_std = torch.tanh(log_std)
        log_std = self.log_std_min + 0.5 * (
          self.log_std_max - self.log_std_min
        ) * (log_std + 1)

        self.outputs['mu'] = mu
        self.outputs['std'] = log_std.exp()

        if compute_pi:
            std = log_std.exp()
            noise = torch.randn_like(mu)
            pi = mu + noise * std
        else:
            pi = None
            entropy = None

        if compute_log_pi:
            log_pi = gaussian_logprob(noise, log_std)
        else:
            log_pi = None

        mu, pi, log_pi = squash(mu, pi, log_pi)

        return mu, log_std

    def log(self, L, step, log_freq=LOG_FREQ):
        if step % log_freq != 0:
            return

        for k, v in self.outputs.items():
            L.log_histogram('train_actor/%s_hist' % k, v, step)

        L.log_param('train_actor/fc1', self.trunk[0], step)
        L.log_param('train_actor/fc2', self.trunk[2], step)
        L.log_param('train_actor/fc3', self.trunk[4], step)


class QFunction(nn.Module):
    """MLP for q-function."""

    def __init__(self, obs_dim, action_dim, hidden_dim):
        super().__init__()

        self.trunk = nn.Sequential(
            nn.Linear(obs_dim + action_dim, hidden_dim), nn.ReLU(),
            nn.Linear(hidden_dim, hidden_dim), nn.ReLU(),
            nn.Linear(hidden_dim, 1)
        )

    def forward(self, obs, action):
        assert obs.size(0) == action.size(0)

        obs_action = torch.cat([obs, action], dim=1)
        return self.trunk(obs_action)


class Critic(nn.Module):
    """Critic network, employes two q-functions."""

    def __init__(
      self, obs_shape, action_shape, hidden_dim, encoder_type,
      encoder_feature_dim, num_layers, num_filters
    ):
        super().__init__()

        self.encoder = make_encoder(
            encoder_type, obs_shape, encoder_feature_dim, num_layers,
            num_filters, output_logits=True
        )

        self.Q1 = QFunction(
            self.encoder.feature_dim, action_shape[0], hidden_dim
        )
        self.Q2 = QFunction(
            self.encoder.feature_dim, action_shape[0], hidden_dim
        )

        self.outputs = dict()
        self.apply(weight_init)

    def forward(self, obs, action, detach_encoder=False):
        # detach_encoder allows to stop gradient propogation to encoder
        obs = self.encoder(obs, detach=detach_encoder)

        q1 = self.Q1(obs, action)
        q2 = self.Q2(obs, action)

        self.outputs['q1'] = q1
        self.outputs['q2'] = q2

        return q1, q2

    def forward_from_feature(self, feature, action):
        # detach_encoder allows to stop gradient propogation to encoder

        q1 = self.Q1(feature, action)
        q2 = self.Q2(feature, action)

        self.outputs['q1'] = q1
        self.outputs['q2'] = q2

        return q1, q2

    def log(self, L, step, log_freq=LOG_FREQ):
        if step % log_freq != 0:
            return

        self.encoder.log(L, step, log_freq)

        for k, v in self.outputs.items():
            L.log_histogram('train_critic/%s_hist' % k, v, step)

        for i in range(3):
            L.log_param('train_critic/q1_fc%d' % i, self.Q1.trunk[i * 2], step)
            L.log_param('train_critic/q2_fc%d' % i, self.Q2.trunk[i * 2], step)


class CURL(nn.Module):
    """
    CURL
    """

    def __init__(self, obs_shape, z_dim, batch_size, critic, critic_target, output_type="continuous"):
        super(CURL, self).__init__()
        self.batch_size = batch_size

        self.encoder = critic.encoder

        self.encoder_target = critic_target.encoder

        self.W = nn.Parameter(torch.rand(z_dim, z_dim))
        self.output_type = output_type

    def encode(self, x, detach=False, ema=False):
        """
        Encoder: z_t = e(x_t)
        :param x: x_t, x y coordinates
        :return: z_t, value in r2
        """
        if ema:
            with torch.no_grad():
                z_out = self.encoder_target(x)
        else:
            z_out = self.encoder(x)

        if detach:
            z_out = z_out.detach()
        return z_out

    def compute_logits(self, z_a, z_pos):
        """
        Uses logits trick for CURL:
        - compute (B,B) matrix z_a (W z_pos.T)
        - positives are all diagonal elements
        - negatives are all other elements
        - to compute loss use multiclass cross entropy with identity matrix for labels
        """
        Wz = torch.matmul(self.W, z_pos.T)  # (z_dim,B)
        logits = torch.matmul(z_a, Wz)  # (B,B)
        logits = logits - torch.max(logits, 1)[0][:, None]
        return logits


class CurlSacAgent(object):
    """CURL representation learning with SAC."""

    def __init__(
      self,
      obs_shape,
      action_shape,
      device,
      args,
      hidden_dim=256,
      discount=0.99,
      init_temperature=0.01,
      alpha_lr=1e-3,
      alpha_beta=0.9,
      alpha_fixed=False,
      actor_lr=1e-3,
      actor_beta=0.9,
      actor_log_std_min=-10,
      actor_log_std_max=2,
      actor_update_freq=2,
      critic_lr=1e-3,
      critic_beta=0.9,
      critic_tau=0.005,
      critic_target_update_freq=2,
      encoder_type='pixel',
      encoder_feature_dim=50,
      encoder_lr=1e-3,
      encoder_tau=0.005,
      num_layers=4,
      num_filters=32,
      cpc_update_freq=1,
      log_interval=100,
      detach_encoder=False,
      curl_latent_dim=128
    ):
        self.args = args
        self.device = device
        self.discount = discount
        self.critic_tau = critic_tau
        self.encoder_tau = encoder_tau
        self.actor_update_freq = actor_update_freq
        self.critic_target_update_freq = critic_target_update_freq
        self.cpc_update_freq = cpc_update_freq
        self.log_interval = log_interval
        self.image_size = obs_shape[-1]
        self.curl_latent_dim = curl_latent_dim
        self.detach_encoder = detach_encoder
        self.encoder_type = encoder_type
        self.alpha_fixed = alpha_fixed

        self.actor = Actor(
            obs_shape, action_shape, hidden_dim, encoder_type,
            encoder_feature_dim, actor_log_std_min, actor_log_std_max,
            num_layers, num_filters
        ).to(device)

        self.critic = Critic(
            obs_shape, action_shape, hidden_dim, encoder_type,
            encoder_feature_dim, num_layers, num_filters
        ).to(device)

        self.critic_target = Critic(
            obs_shape, action_shape, hidden_dim, encoder_type,
            encoder_feature_dim, num_layers, num_filters
        ).to(device)

        self.critic_target.load_state_dict(self.critic.state_dict())

        # tie encoders between actor and critic, and CURL and critic
        self.actor.encoder.copy_conv_weights_from(self.critic.encoder)

        self.log_alpha = torch.tensor(np.log(init_temperature)).to(device)
        self.log_alpha.requires_grad = True
        # set target entropy to -|A|
        self.target_entropy = -np.prod(action_shape)

        # optimizers
        self.actor_optimizer = torch.optim.Adam(
            self.actor.parameters(), lr=actor_lr, betas=(actor_beta, 0.999)
        )

        self.critic_optimizer = torch.optim.Adam(
            self.critic.parameters(), lr=critic_lr, betas=(critic_beta, 0.999)
        )
        self.log_alpha_optimizer = torch.optim.Adam(
            [self.log_alpha], lr=alpha_lr, betas=(alpha_beta, 0.999)
        )

        if self.args.lr_decay is not None:
            # Actor is halved due to delayed update
            self.actor_lr_scheduler = torch.optim.lr_scheduler.MultiStepLR(self.actor_optimizer, milestones=np.arange(15, 150, 15) * 5000, gamma=0.5)
            self.critic_lr_scheduler = torch.optim.lr_scheduler.MultiStepLR(self.critic_optimizer, milestones=np.arange(15, 150, 15) * 10000, gamma=0.5)

        if self.encoder_type == 'pixel':
            # create CURL encoder (the 128 batch size is probably unnecessary)
            self.CURL = CURL(obs_shape, encoder_feature_dim,
                             self.curl_latent_dim, self.critic, self.critic_target, output_type='continuous').to(self.device)

            # optimizer for critic encoder for reconstruction loss
            self.encoder_optimizer = torch.optim.Adam(
                self.critic.encoder.parameters(), lr=encoder_lr
            )

            self.cpc_optimizer = torch.optim.Adam(
                self.CURL.parameters(), lr=encoder_lr
            )
        self.cross_entropy_loss = nn.CrossEntropyLoss()

        self.train()
        self.critic_target.train()

    def train(self, training=True):
        self.training = training
        self.actor.train(training)
        self.critic.train(training)
        if self.encoder_type == 'pixel':
            self.CURL.train(training)

    @property
    def alpha(self):
        return self.log_alpha.exp()

    def select_action(self, obs):
        with torch.no_grad():
            if not isinstance(obs, torch.Tensor):
                obs = torch.from_numpy(obs)
            obs = obs.to(torch.float32).to(self.device)
            obs = obs.unsqueeze(0)
            mu, _, _, _ = self.actor(
                obs, compute_pi=False, compute_log_pi=False
            )
            return mu.cpu().data.numpy().flatten()

    def sample_action(self, obs):
        if obs.shape[-1] != self.image_size:
            obs = utils.center_crop_image(obs, self.image_size)

        with torch.no_grad():
            if not isinstance(obs, torch.Tensor):
                obs = torch.from_numpy(obs)
            obs = obs.to(torch.float32).to(self.device)
            obs = obs.unsqueeze(0)
            mu, pi, _, _ = self.actor(obs, compute_log_pi=False)
            return pi.cpu().data.numpy().flatten()

    def update_critic(self, obs, action, reward, next_obs, not_done, L, step):
        with torch.no_grad():
            _, policy_action, log_pi, _ = self.actor(next_obs)
            target_Q1, target_Q2 = self.critic_target(next_obs, policy_action)
            target_V = torch.min(target_Q1,
                                 target_Q2) - self.alpha.detach() * log_pi
            target_Q = reward + (not_done * self.discount * target_V)
            # comp_Q = reward + (not_done * self.discount * torch.min(target_Q1, target_Q2))
            # comp_E = reward + (not_done * self.discount * (- self.alpha.detach() * log_pi))

        # get current Q estimates
        current_Q1, current_Q2 = self.critic(
            obs, action, detach_encoder=self.detach_encoder)
        critic_loss = F.mse_loss(current_Q1,
                                 target_Q) + F.mse_loss(current_Q2, target_Q)
        if step % self.log_interval == 0:
            L.log('train_critic/loss', critic_loss, step)
            # L.log('train/q1', torch.mean(current_Q1), step)
            # L.log('train/comp_Q', torch.mean(comp_Q), step)
            # L.log('train/comp_E', torch.mean(comp_E), step)

            # critic_stats = get_optimizer_stats(self.critic_optimizer)
            # for key, val in critic_stats.items():
            #     L.log('train/critic_optim/' + key, val, step)

        # Optimize the critic
        self.critic_optimizer.zero_grad()
        critic_loss.backward()
        self.critic_optimizer.step()

        if self.args.lr_decay is not None:
            self.critic_lr_scheduler.step()
            L.log('train/critic_lr', self.critic_optimizer.param_groups[0]['lr'], step)


        self.critic.log(L, step)

    def update_actor_and_alpha(self, obs, L, step):
        # detach encoder, so we don't update it with the actor loss
        _, pi, log_pi, log_std = self.actor(obs, detach_encoder=True)
        actor_Q1, actor_Q2 = self.critic(obs, pi, detach_encoder=True)

        actor_Q = torch.min(actor_Q1, actor_Q2)
        actor_loss = (self.alpha.detach() * log_pi - actor_Q).mean()

        if step % self.log_interval == 0:
            L.log('train_actor/loss', actor_loss, step)
            L.log('train_actor/target_entropy', self.target_entropy, step)
        entropy = 0.5 * log_std.shape[1] * \
                  (1.0 + np.log(2 * np.pi)) + log_std.sum(dim=-1)
        if step % self.log_interval == 0:
            L.log('train_actor/entropy', entropy.mean(), step)

        # optimize the actor
        self.actor_optimizer.zero_grad()
        actor_loss.backward()
        self.actor_optimizer.step()

        if self.args.lr_decay is not None:
            self.actor_lr_scheduler.step()
            L.log('train/actor_lr', self.actor_optimizer.param_groups[0]['lr'], step)


        self.actor.log(L, step)

        # if step % self.log_interval == 0:
        #     actor_stats = get_optimizer_stats(self.actor_optimizer)
        #     for key, val in actor_stats.items():
        #         L.log('train/actor_optim/' + key, val, step)


        if not self.alpha_fixed:
            self.log_alpha_optimizer.zero_grad()
            alpha_loss = (self.alpha *
                          (-log_pi - self.target_entropy).detach()).mean()
            if step % self.log_interval == 0:
                L.log('train_alpha/loss', alpha_loss, step)
                L.log('train_alpha/value', self.alpha, step)
            alpha_loss.backward()
            self.log_alpha_optimizer.step()

    def update_cpc(self, obs_anchor, obs_pos, cpc_kwargs, L, step):

        z_a = self.CURL.encode(obs_anchor)
        z_pos = self.CURL.encode(obs_pos, ema=True)

        logits = self.CURL.compute_logits(z_a, z_pos)
        labels = torch.arange(logits.shape[0]).long().to(self.device)
        loss = self.cross_entropy_loss(logits, labels)

        self.encoder_optimizer.zero_grad()
        self.cpc_optimizer.zero_grad()
        loss.backward()

        self.encoder_optimizer.step()
        self.cpc_optimizer.step()
        if step % self.log_interval == 0:
            L.log('train/curl_loss', loss, step)

    def update(self, replay_buffer, L, step):
        if self.encoder_type == 'pixel':
            obs, action, reward, next_obs, not_done, cpc_kwargs = replay_buffer.sample_cpc()
        else:
            obs, action, reward, next_obs, not_done = replay_buffer.sample_proprio()

        if step % self.log_interval == 0:
            L.log('train/batch_reward', reward.mean(), step)

        start_time = time.time()
        self.update_critic(obs, action, reward, next_obs, not_done, L, step)
        # print('critic update time:', time.time() - start_time)
        if step % self.actor_update_freq == 0:
            start_time = time.time()
            self.update_actor_and_alpha(obs, L, step)
            # print('actor update time:', time.time() - start_time)

        if step % self.critic_target_update_freq == 0:
            utils.soft_update_params(
                self.critic.Q1, self.critic_target.Q1, self.critic_tau
            )
            utils.soft_update_params(
                self.critic.Q2, self.critic_target.Q2, self.critic_tau
            )
            utils.soft_update_params(
                self.critic.encoder, self.critic_target.encoder,
                self.encoder_tau
            )

        if step % self.cpc_update_freq == 0 and self.encoder_type == 'pixel':
            start_time = time.time()
            obs_anchor, obs_pos = cpc_kwargs["obs_anchor"], cpc_kwargs["obs_pos"]
            self.update_cpc(obs_anchor, obs_pos, cpc_kwargs, L, step)

    def save(self, model_dir, step):
        torch.save(
            self.actor.state_dict(), '%s/actor_%s.pt' % (model_dir, step)
        )
        torch.save(
            self.critic.state_dict(), '%s/critic_%s.pt' % (model_dir, step)
        )
        
        if self.encoder_type == 'pixel':
            self.save_curl(model_dir, step)

    def save_curl(self, model_dir, step):
        torch.save(
            self.CURL.state_dict(), '%s/curl_%s.pt' % (model_dir, step)
        )

    def load(self, model_dir, step):
        self.actor.load_state_dict(
            torch.load('%s/actor_%s.pt' % (model_dir, step))
        )
        self.critic.load_state_dict(
            torch.load('%s/critic_%s.pt' % (model_dir, step))
        )
        
        if self.encoder_type == 'pixel':    
            self.CURL.load_state_dict(
                torch.load('%s/curl_%s.pt' % (model_dir, step))
            )